Novel pathways for oxygen insertion into unactivated C-H bonds by dioxiranes. Transition structures for stepwise routes via radical pairs and comparison with the concerted pathway.

The oxygen insertion into C-H bonds (of methane, isobutane, and acetone) by dioxiranes (parent dioxirane and dimethyldioxirane) to give alcohols was studied with the DFT theory, using both restricted and unrestricted B3LYP methods, and 6-31G(d) and 6-311+G(d,p) basis sets to evaluate the feasibility of stepwise mechanisms and their competition with the concerted counterpart. Confirming previous results by other authors, we have located, with the RB3LYP method, concerted TSs in which the oxygen bound to be inserted interacts very strongly with the hydrogen atom and very weakly with the carbon atom of the C-H bond. These TSs nicely explain all the experimental observations (e.g., configuration retention at the chiral centers), but all of them exhibit an RHF --> UHF wave function instability that preclude considering them as genuine transition structures. We also were able to characterize, with UB3LYP methods, two alternative two-step processes that can lead to final products (alcohol + carbonyl compound) via singlet radical pair intermediates. For the first step of both processes we located genuine diradicaloid TSs, namely, TSs rad,coll and TSs rad,perp, that have stable wave functions. In TSs rad,coll the alkane C-H bond tends to be collinear with the breaking O(1)- - -O(2) bond while in TSs rad,perp the alkane C-H bond is almost perpendicular to the O(1)- - -O(2) bond. The first step, of both processes, can represent an example of a "molecule induced homolysis" reaction: collision between alkane and dioxirane brings about the homolytic cleavage of the dioxirane O-O bond and the hydrogen abstraction follows afterward to produce the diradicaloid TS that then falls down to a singlet radical pair. This hypothesis was fully confirmed by IRC analysis in the case of TSs rad,coll. The possible pathways that lead from the intermediate radical pair to final products are discussed as well as the hypothesis that the radical collinear TSs may collapse directly to products in a "one-step nonconcerted" process. However, diradical mechanisms cannot explain the experimental data as satisfactorily as the concerted pathway does. As for computational predictions about competition of diradical vs concerted mechanisms, they strongly depend (i) on the alkane C-H type, (ii) on whether gas phase or solution is considered, and (iii) on the basis set used for calculations. In short, the concerted TS benefits, with respect to the corresponding diradicaloid TSs, of alkyl substitution at the C-H center, solvation effects, and basis set extension. Actually, in the case of DMD reactions with methane and acetone, the diradicaloid TSs are always (both in gas phase and in solution and with both the basis sets used) strongly favored over their concerted counterpart. In the case of DMD reaction with isobutane tertiary C-H bond the large favor for the diradicaloid TSs over the concerted TS, predicted in gas phase by the B3LYP/6-31G(d) method, progressively decreases as a result of basis set extension and introduction of solvent effects: the higher theory level [B3LYP/6-311+G(d,p)] suggests that in acetone solution TS conc has almost the same energy as TS rad,perp while TS rad,coll resides only 2 kcal/mol higher.